Efforts to refine therapy have reduced the mortality rate and increased the organ preservation rate in patients with head and neck cancer during the past five years. Further, improvement in therapy outcome would result from identification of biological markers that can predict the probabilities would respond to different therapy modalities. Such an achievement would make it possible to select the best therapy for individual patients based on their distinct tumor features. One potential marker is epidermal growth factor receptor (EGFR), a transmembrane glycoprotein with an intracellular domain possessing intrinsic tyrosine kinase activity. The marker is over-expressed in many tumors including the vast majority of head and neck carcinomas. Emerging data, mostly from in vitro studies, suggest an association between extent of EGFR over-expression and resistance to some cytotoxic drugs and radiation. Preliminary evidence linking EGFR over- expression and poor relapse-free and overall survival has also been reported. The validity of these interesting leads has not, however, been established in a sufficiently large, well-characterized patient population receiving consistent therapy. We propose to identify correlations between EGFR level and clinical outcome in 973 patients with stage III and IV head and neck carcinomas enrolled in a prospective randomized trial of Radiation Therapy Oncology Group. The original study was designed to assess the relative efficacy of four radiation fractionation regimens. Since treatment consisted of radiotherapy only in all patients, the study population is ideal for addressing the value of EGFR expression in predicting tumor response to radiotherapy in general and to altered fractionation regimens, particularly accelerated fractionations. Biopsy samples and follow-up data are available for these patients. We hope to find a method for identify patients with tumors most likely to respond to radiotherapy so that only those with radio-resistant tumors would be subjected to extensive surgery and its attendant morbidity. Our specific aims are as follows: 1) define the range and distribution of EGFR protein expression in stage III and IV head and neck carcinomas; 2) determine the potential association between EGFR expression and clinical prognostic factors and other biomarkers; 3) test a hypothesis that the basal EGFR expression level is an independent prognostic indicator for local-regional control of stage III & IV head and neck cancer treated with radiotherapy alone; 4) test a second hypothesis that the basal EGFR expression level is positively correlated with the extent of in vivo tumor clonogen proliferation during the course of radiotherapy; and 5) assess whether the addition of EGFR expression level is positivity correlated with the extent of in vivo tumor clonogen proliferation during the course of radiotherapy; and 5) assess whether the addition of EGFR expression level imposes risk assessment by RTOG's recursive partitioning analysis had hence refines patient stratification in future trials.